Segregation of chlorophyll a incorporated into lipid bilayers

D. Klee, K. W. Seo, Y. S. Kang. Photoinduced electron transfer of ..... DOI: 10.1016/0014-5793(76)80094-4. Paul-André Siegenthaler, Antoine TrémoliÃ...
0 downloads 0 Views 715KB Size
CHLOROPHYLL IN LIPID BILAYERS

Segregation of Chlorophyll a Incorporated into Lipid Bilayers? A. G. Lee

ABSTRACT: Absorption and fluorescence spectra are reported for chlorophyll a incorporated into a number of aqueous phospholipid dispersions. Absorption spectra show that in dipalmitoylphosphatidylcholine bilayers, monomeric and oligomeric forms of chlorophyll a are present in both the gel and liquid crystalline phases. The formation of aggregates of chlorophyll a is reflected in the fluorescence spectra by a marked concentration quenching. In bilayers containing small proportions of chlorophyll a , a marked increase in aggregation occurs a t the transition temperatures

that can be detected calorimetrically. At higher concentrations (>1 chlorophyl1:lOO lipid), the “pretransition” is abolished in the phosphatidylcholines, and the main transition is broadened, consistent with an orientation for the chlorophyll a with the chlorine ring in the head group region and the phytol chain in the fatty acid chain region of the bilayer. In mixtures of saturated and unsaturated lipids, there is no preferential segregation of the chlorophyll a into the unsaturated lipid.

D e s p i t e its obvious importance for any study of photosynthesis, relatively little is known about the organization of chlorophylls incorporated into lipid bilayers. This is in marked contrast to the very considerable amount of information available concerning the state of the chlorophylls in organic solvents (Vernon and Seely, 1966; Cotton et al., 1974) and in monolayers (Ke, 1966). However, in the light of the marked aggregation of chlorophyll a found in nonpolar solvents and the similarity between the absorption spectra of many of these aggregates and chlorophyll a in photosynthetic membranes (Cotton et al., 1974), it has often been suggested that aggregates of chlorophyll a are present in photosynthetic membranes. Recent studies have established some of the characteristics of component segregation within lipid bilayers (Lee, 1975). In particular, studies of phospholipid-phospholipid (Shimshick and McConnell, 1973) and phospholipid-steroid mixtures (Trauble and Sackmann, 1972) have established the presence of segregated pools of lipid and steroid in bilayers a t temperatures below that of the gel to liquid crystalline phase transition of one or more of the lipid components of the mixture. However, phospholipids in the liquid crystalline phase seem to be completely miscible in all proportions (Shimshick and McConnell, 1973), and the spin-labeled steroid analog studied by Trauble and Sackmann (1972) is completely miscible with lipid in the liquid crystalline phase a t least up to a steroid/lipid molar ratio of 1:4. Since the lipids of most photosynthetic plants contain a relatively high proportion of polyunsaturated fatty acids (Galliard, 1973), the photosynthetic membrane will be predominantly in the liquid crystalline phase a t ambient temperature. Aggregation of chlorophyll a within the membrane could then only occur either because of specific chlorophyll-chlorophyll interaction, or because the chlorophyll was in some way tightly coupled to the protein component of the membrane. In this paper are reported experiments which show that aggregates of chlorophyll a are present in lipid bilayers, both when the lipid is in the gel and in the liquid crystalline

phases, although the proportion of aggregated chlorophyll is increased a t the expense of monomer when the lipid is in the gel state. The formation of aggregates of chlorophyll a in lipid bilayers in the liquid crystalline state shows that chlorophyll-chlorophyll interactions alone are sufficient to explain their formation, and thus the formation of “antenna” chlorophyll in the photosynthetic membrane.

From the Department of Physiology and Biochemistry, University of Southampton, Southampton SO9 3TU. Receiaed M a y 8, 1975.

Materials and Methods

Dipalmitoylphosphatidylcholine was obtained from Koch-Light, dioleoylphosphatidylcholine from P-L Biochemicals, and dimyristoylphosphatidylethanolaminefrom Fluka. Chlorophyll a was purified by column chromatography on powdered sugar columns by the method of Strain and Svec (1946). Chlorophyll a concentrations were estimated from absorption spectra, using the extinction coefficients given by Strain and Svec (1966). Lipids plus chlorophyll a (1.6 X mol) dissolved in chloroform were mixed in 10-ml stoppered flasks and evaporated to dryness under a stream of nitrogen. Buffer (4 ml; 0.01 M Tris-HC1 (pH 7.2)-0.1 M NaC1) was added and the mixture shaken on a Vortex mixer. The optical density a t 670 nm was less than 0.1 for all samples. Absorption spectra were recorded on a Cary 14 spectrophotometer. Fluorescence measurements were made on an Aminco Bowman SPF fluorimeter. Samples were continuously stirred during the fluorescence experiments, and temperatures were measured by a thermocouple. Fluorescence of chlorophyll a was excited a t 420 nm and recorded a t 670 nm. Over the lipid/chlorophyll a ratios used in these experiments (5OOO:l to 30:1), the absorption intensity a t 670 nm did not change significantly, so that changes in fluorescence intensity correspond to changes in fluorescence quantum yield. Results Absorption spectra of chlorophyll a in dioleoylphosphatidylcholine and dipalmitoylphosphatidylcholine bilayers a t 45OC and a t chlorophyll/lipid molar ratios of 1:800 and 1.8, respectively, are shown in Figure l a and b. The maximum of the chlorophyll a a peak appears a t 668-672 nm, depending on concentration, and the spectra are virtually independent of temperature in the range 10-50°C. With inBIOCHEMISTRY, VOL

14, NO

20, 1975

4397

’ Abs. ,

4-

a

Wavelength (nm) I : Absorption spectrum of chlorophyll a (1.6 X mol) in liposomes o f (a) dioleoylphosphatidylcholine at a chlorophyll/lipid molar ratio of 1 :800, (b) dipalmitoylphosphatidylcholine at a chlorophyll/lipid molar ratio of 1:8 (absorption scale X 2 ) , both dispersed in 4 ml. Tris buffer at 45OC. The difference spectrum showing aggregated chlorophyll a formed at high concentration, produced by subtracting the spectrum of Figure l a from that of Figure Ib, after normalizing the latter spectrum to the maximum peak height of the former, is shown in c. FI(;LRF

creasing concentration, there is a marked increase in peak width a t half-height, from ca. 30 nm (up to a chlorophyll/ lipid molar ratio of 1:80), to 37 nm at a molar ratio of 1:40 and to 43 nm a t a molar ratio of 1:8. The increase in peak width cannot be attributed to an artifact such as light scattering, since in these experiments the amount of chlorophyll is kept constant, and the chlorophyll/lipid molar ratio is increased by decreasing the amount of lipid: this means that the samples which show the greatest peak width are those ~ i t the h least amount of lipid, and which therefore scatter light the least. The increase in width seems rather to be due to the appearance of a shoulder a t ca. 685 nm a t the higher concentrations. This is shown most clearly by the difference spectrum (Figure IC). Deconvolution studies of spectra of chlorophyll a in nonpolar organic solvents have previously shown the presence of a major band a t ca. 680 nm, due to aggregated chlorophyll (Cotton et al., 1974). Fluorescence Spectra in Dipalmitoylphosphatidylcholine, In bilayers of dipalmitoylphosphatidylcholine,chlorophyll a has a fluorescence maximum a t 670 nm, whereas in diethyl ether the fluorescence maximum is at 667 nm. I n the lipid bilayer, it is necessary to determine whether or not there are more than one fluorescent chlorophyll a species present. since the absorption spectra of these systems show the presence of chlorophyll aggregates. This can be done since the intensity of fluorescence emitted a t a certain wavelength by a mixture of several fluorescent forms with different positions of absorption bands will depend on the wavelength of excitation. Within experimental error, fluorescence spectra normalized to equal intensity a t 670 nm are identical with excitation wavelengths between 390 and 440 nm, and a t temperatures between 20 and 50OC. This suggests therefore that there is only one fluorescent form of chlorophyll a present in bilayers of dipalmitoylphosphatidylcholine, both in the gel and liquid crystalline states. Figure 2 shows the effect on fluorescence intensity of changing the lipid/chlorophyll a molar ratio, while keeping the amount of chlorophyll a constant. In the absence of any concentration quenching, the fluorescence intensity in this BIOCHEMISTRY.

VOL.

p

z . \

0

4398

3 ‘

1 4 , 10.2 0 , 1 9 7 5

10

15

I

20

1 -

25

F I G U R E 2: Fluorescence intensity (arbitrary units) at 670 nm vs. lipid/ chlorophyll molar ratio 1 :c. The amount of chlorophyll a was kept constant at 1.6 X mol. 0 at 50’C; 0 at 2OOC. Curves 1, 2, and 3 are calculated from a random distribution model (see text): curve I . interaction distance 60 A, lipid cross-sectional area 70 A*: curve 2 , interaction distance 86 A, lipid cross-sectional area 70 A2: curve 3, interaction distance 1 I9 A,lipid cross-sectional area 40 A2.

experiment would remain constant: in fact, the fluorescence intensity is much greater for those liposomes containing a high proportion of lipid. Analogous concentration quenching has been observed for many fluorescent dyes, and attributed to deactivating collisions between excited molecules and molecules in the ground state (Seliger and McEIroy, 1965). In that case, the fluorescence intensity N should follow the Stern-Volmer relation

where No is the number of quanta absorbed per second, 70 is the natural lifetime of the excited state, and q is the number of effective quenching collisions per second. Since q should be proportional to the concentration, eq 1 can be rewritten as NoIN = 1

+ C/C*

(2)

where c is the concentration of fluorophor and c * is the concentration a t which the fluorescence yield has fallen to onehalf. If collisional quenching were important for chlorophyll a in these lipid bilayers, a plot of the reciprocal of fluorescence intensity against chlorophyll/lipid ratio would be a straight line. This is not so for the data plotted in Figure 2. An alternative explanation that has been proposed for the concentration quenching of dyes is the formation of ground state dimers or polymers which are nonfluorescent (Rabinowitch and Epstein, 1941). There are two forms that such quenching could take in our case. I n the first, chlorophyll a molecules are randomly distributed within the lipid bilayer, and two chlorophyll a molecules closer than some given interaction distance form a nonfluorescent pair. I n the second, the distribution of chlorophyll a within the lipid bilayer is nonrandom, and nonfluorescent dimers or oligomers are formed by specific chlorophyll-chlorophyll interaction. If the distribution of chlorophyll molecules is random, then the probability that the nearest neighbor chlorophyll molecule to one particular chlorophyll molecule in the bilayer plane is between r and r d r can be written as w(r)dr, where w(r) is the probability density. This is equal to the probability that there is no chlorophyll closer than r, times the probability that another chlorophyll molecule ex-

+

CHLOROPHYLL

IN LIPID BILAYERS

ists in the circular ring between r and r probability density is given by

+ d r . Thus the

[

w(r) = 1 - xrw(r)dr]2?rm

(3)

Integration then leads to w(r) = 2 ~ r exp(-rr2n) n

(4)

Here n is the number of chlorophyll a molecules in unit area. For simplicity in the calculations, the area occupied by a chlorophyll a molecule in the bilayer was taken to be equal to that of a lipid molecule: with the very low proportions of chlorophyll used in these experiments, the fact that the cross-sectional area of a phytol chain is ca. one-half of that of a lipid molecule will have no significant effect. Figure 2 shows attempts to fit the experimental data to this random distribution model, the calculations being normalized to the observed fluorescence intensity a t a chlorophyll/ lipid molar ratio of 2 X 10-4:l. In the liquid crystalline phase, a lipid cross-sectional area of 70 A2 was assumed (Lee, 1975), and the best fit to the data was then obtained with a n interaction distance of 86 A, although the calculated fluorescence for bilayers containing little lipid was then rather too low (curve 2, Figure 2). The data for the high chlorophyll/lipid ratios a t 5OoC were better fitted with an interaction distance of 60 8, (curve 1, Figure 2) but then the fluorescence intensities calculated a t low chlorophyll/lipid ratios were too high. Decreasing the area occupied per lipid molecule increases the calculated proportion of nonfluorescent chlorophyll a t any given chlorophyll/lipid ratio and interaction distance, without, however, changing the shape of the fluorescence-molar ratio curve. In the gel state, the surface area of a lipid molecule is ca. 40 A2 (Lee, 1975), and curve 3 of Figure 2 shows that a good fit to the fluorescence data a t 2OoC in the more dilute bilayers can then be obtained with a n interaction distance of 119 A. All of these interaction distances seem to be unreasonably large to represent the formation of nonfluorescent pairs of chlorophyll a molecules. The area of the porphyrin group of a chlorophyll is ca. 225 A2, and if, as has been suggested, the porphyrin ring is tilted a t an angle of ca. 50° to the bilayer plane (Hoff, 1974), then the area occupied a t the bilayer surface is ca. 144 A*. This means that the centers of two porphyrin rings must be closer than ca. 16 8, to allow overlap, and in crystals of ethyl chlorophyllide a dihydrate, the Mg-Mg distance between adjacent molecules can be estimated from the published data (Strouse, 1974) to be less than 10 8,. Since the interaction distances calculated from the random distribution model are considerably greater than those expected from the size of the chlorophyll a molecule, it seems that the distribution of chlorophyll a molecules is not random, but rather that nonfluorescent dimers or oligomers are formed. If only monomeric and dimeric chlorophyll a are present in the bilayer, then if [Chl], [Chlz], and m represent the mole fractions of monomer, dimer, and total chlorophyll, it follows that m = [Chl]

+ 2[Ch12]

If the equilibrium constant for dimerization is K, then

(5)

V I

0

I

5

I

10

15

20 c*1@

25

FIGURE 3: Attempted fit of fluorescence vs. concentration data to a dimer model. Data as in Figure 4. Curve 1, K = 500 (mole fraction)-'; curve 2, K = 1000 (mole fraction)-'; curve 3, K = 3800 (mole fraction)-'.

Thus, [Chl] = (IhK)[(8Km

+ I ) ' / * - 13

(7)

If the dimer is nonfluorescent, then the fluorescence intensity a t any given chlorophyll/lipid ratio is proportional to the amount of monomeric chlorophyll present. Figure 3 shows attempts to fit the fluorescence data for chlorophyll a in dipalmitoylphosphatidylcholine a t 5OoC to such a model of dimer formation: again, the calculations were normalized to the observed fluorescence intensity a t a chlorophyll/lipid molar ratio of 2 X 10-4: 1. In the most dilute bilayers a reasonable fit can be obtained with an equilibrium constant for dimerization of K = 500 (mole fraction)-': assuming a density of 1 for the lipid (Levine, 1973), this corresponds to K 600 1. mol-'. At chlorophyll/lipid molar ratios of greater than 1.3 X 10-3:1, however, the fit is very poor. Increasing the equilibrium constant improves the fit a t higher chlorophyll/lipid molar ratios, but then predicts a lower than observed fluorescence a t low chlorophyll/lipid molar ratios. It seems therefore that although dimers may be formed a t low chlorophyll concentrations, aggregates of more than two chlorophylls are certainly present a t chlorophyll/lipid molar ratios greater than ca. 2 X 10-3:1, when the lipid is in the liquid crystalline state. When the lipid is in the gel state, aggregation beyond the dimer would occur a t even lower chlorophyll/lipid ratios. It is also possible that some of the fluorescence quenching observed a t higher chlorophyll concentrations could be due to nonradiative energy transfer to the nonfluorescent dimers of chlorophyll. However, estimated values for the rate of nonradiative transfer suggest that this can only be a very minor effect. The pairwise energy transfer rate 7 by dipoledipole resonance interaction is given by the Forster theory as (Seliger and McElroy, 1965)

where 70 is the natural fluorescence lifetime, R is the separation between the pair of chlorophyll molecules, and Ro is the critical interaction distance. The values of Ro and 70 have been estimated as 65-70 8, and 15 nsec, respectively (Duysens, 1964). At a chlorophyll/lipid molar ratio of 2.3 X 10-3:1, the average separation of chlorophyll molecules is 170 A, and the pairwise transfer rate is then 2 X lo5 sec-I, so that the probability of excitation energy transfer to a chlorophyll dimer, even if it were the nearest neighbor, BIOCHEMISTRY. VOL.

1 4 , N O . 2 0 , 197.5

4399

LEE

i

4

DMPE

A-

31

3 7

1

24 0: 10

I

20

30

40 T(OC)

sb

6b

/ J

li

4: Fluorescence intensity of chlorophyll a vs. temperature, in liposomes of dioleoylphosphatidylcholine(DOL), dipalmitoylphosphatidylcholine (DPL), and dimyristoylphosphatidylethanolamine (DMPE), at a chlorophyll/lipid molar ratio of 1:430. FIGURE

would be very small. At a chlorophyll/lipid molar ratio of 2.6 X 10-2:1, the pairwise transfer rate has increased to 3 X los sec-', but even then the probability of transfer to a nonfluorescent chlorophyll dimer will be very small. Effect of Temperature on Fluorescence. Although temperature has no effect on the wavelength of fluorescence emission, there is a marked decrease in fluorescence intensity when the temperature drops below that of the liquid crystalline to gel phase transition. This is shown by the data in Figure 4, for chlorophyll a present at a chlorophyll/lipid molar ratio of 1 :430 in bilayers of dioleoylphosphatidylcholine, dipalmitoylphosphatidylcholine,and dimyristoylphosphatidylethanolamine. The temperature of the gel to liquid crystalline phase transition for dioleoylphosphatidylcholine is -22OC (Phillips et al., 1972), and there is relatively little effect of temperature on the fluorescence of chlorophyll a in this lipid. There is a slight decrease in fluorescence intensity below ca. 25OC, which could correspond to the reported onset of cluster formation in dioleoylphosphatidylcholinea t ca. 3OoC (Lee et al., 1974): although the change in fluorescence intensity below 25OC is only slight, it is in contrast to the usual observation of an increase in fluorescence as the temperature is lowered, due to a decrease in the rates of collisional quenching and other processes (Guilbault, 1973). In bilayers of dipalmitoylphosphatidylcholinea t a chlorophyll/lipid molar ratio of 1:430, there is a marked drop in fluorescence intensity a t 40.5OC. This agrees well with the phase transition temperature of 41.75OC found by calorimetry (Hinz and Sturtevant, 1972) and agrees exactly with that detected by the change in the partitioning of the small molecule Tempo' (I) (Metcalfe et al., 1972; Shimshick and McConnell, 1973). The reduction of fluorescence intensity for chlorophyll a at this temperature can be interpreted as a partial exclusion of the freely dispersed, monomeric chlorophyll from the crystalline lipid region, with the formation of segregated regions of nonfluorescent (aggregated) chloro-

>r)